Announcement

Collapse
No announcement yet.

An Open-Source Bootloader For Windows Lets You Run Off Btrfs, Other Possibilities

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • oiaohm
    replied
    Originally posted by maharmstone View Post
    The author here. Oiaohm, you're talking rubbish. Quite apart from anything else, "BTRFS_SEND_CMD_SUBVOL" has only ever been used in my code - see https://www.google.com/search?q=BTRFS_SEND_CMD_SUBVOL. You'll see that that value comes from the btrfs.h that's shared with WinBtrfs, which uses it fully.
    The problem is the list of values there align what btrfs progs have used. This could be a case of bad luck. But file has lists of lots of things with no comments on how this use used or acquired. So this leaves open to interpretation where it came from. But you do have a acquirement note at the top.

    https://github.com/maharmstone/quibb...rfs/btrfs.h#L3
    This is another problem.
    https://btrfs.wiki.kernel.org/index.php/On-disk_Format
    The bad news that is a GPLv2 file by default by Oracles current arguments if they win. And you have already directly admitted to be derivative work from it.

    Originally posted by maharmstone View Post
    The court case is about whether documentation for proprietary software is in itself proprietary. It's a bit of a leap from that to claim that,
    No read the Oracle case closer. The openjdk documentation is LGPL and google has made Apache 2.0 licensed implementation.

    So this is not proprietary vs proprietary. This is Open Source License conflict of LGPL vs Apache 2.0 where does the line of derivative work in fact exist.

    You need to be watching the Google vs Oracle case carefully.

    Originally posted by maharmstone View Post
    a) this means documentation about something open-source is proprietary,
    If Oracle wins documentation about open-source work is License encumbered and derivative work issue can come from the documentation or the code. Currently most derivative work cases have been about code to code not code made from documentation. This is why Google and Oracle are in court there was no prior cases to settle their dispute with.

    Originally posted by maharmstone View Post
    b) even if it were, Oracle would choose to enforce this,
    The Google vs Oracle case is Oracle attempting to enforce this. Question is will they enforce this on you who knows they could once they have ruling. Second time around is a lot cheaper and faster for Oracle if Oracle has won because they will have precedent on their side.

    Originally posted by maharmstone View Post
    c) even if they did, Linus wouldn't remove their code from the kernel because they've violated the principles of the GPL.
    No Linus will do nothing. Linux kernel is only GPLv2 does not have a or latter clause. Oracle deciding to enforce GPLv2 on you because you used GPLv3 without a or latter clause it not violating Linus believe of GPL. In fact using GPLv2 code in a GPLv3 work when the GPLv2 code does not contain a or latter clause is a GPLv2 license violation..

    Instead would be making sure your code could be taken back into mainline Linux kernel in future. Linus the head of the Linux Kernel project has no reason to be upset if Oracle goes after some upstart who GPLv3 items that are Linux kernel derivative that should be GPLv2. So thinking Linus will save you is being foolish here.

    Originally posted by maharmstone View Post
    And even if what you'd said was right, I'd just put a notice at the top saying the code's not legal to use in the United States.
    That also means you need to remove yourself from github and cannot use gitlab either. USA hosting servers are not allowed to host copyright infringement items.

    maharmstone basically you have not been careful enough. What you have done is legal if Oracle loses against Google. If Oracle wins what you have done will be illegal in the USA and many countries with treaties with the USA over copyright.

    People implementing things have been ignoring what the license of the source documentation is for a long time Google vs Oracle case throws this on it ear as in API documentation copyright does effect anything based off of it instead of the past believe that API and it documentation copyright was a instant dead end that derivative work stuff of copyright law did not apply.

    Like it or not you have not commented your code well enough so have left sections of the header file open to possible wrong interpenetration.

    You have also ignored the source documentation copyright that has been normal to-do that if Oracle wins is legally wrong in the USA courts and treaty linked countries.

    Please note if API documentation copyright can be derivative work issue has not been taken to court in any other country other than the USA at this stage so the country where you are in time may or may not agree with the USA in latter cases.

    Basically Google vs Oracle case if Oracle wins does threaten to disrupt the landscape. If you are accepting code from third parties until the Google vs Oracle case is sorted out would pay to ask them if putting their work under GPLv2.0 same as Linux kernel be acceptable if Google loses.

    Basically simplest and safest solution is WinBtrfs and Quibble end up licensed under GPLv 2.0 as the Linux kernel. If not you need to watch the Google vs Oracle case closely because depending how this goes you could have a licensing issue.

    Leave a comment:


  • maharmstone
    replied
    The author here. Oiaohm, you're talking rubbish. Quite apart from anything else, "BTRFS_SEND_CMD_SUBVOL" has only ever been used in my code - see https://www.google.com/search?q=BTRFS_SEND_CMD_SUBVOL. You'll see that that value comes from the btrfs.h that's shared with WinBtrfs, which uses it fully.

    The court case is about whether documentation for proprietary software is in itself proprietary. It's a bit of a leap from that to claim that, a) this means documentation about something open-source is proprietary, b) even if it were, Oracle would choose to enforce this, and c) even if they did, Linus wouldn't remove their code from the kernel because they've violated the principles of the GPL. Please don't spread FUD.

    And even if what you'd said was right, I'd just put a notice at the top saying the code's not legal to use in the United States.

    Leave a comment:


  • oiaohm
    replied
    Originally posted by pal666 View Post
    here is your problem. filesystem isn't an api. it's data structure. data structures aren't copyrightable. in api you have to use same names, otherwise client code wouldn't compile. in data structure you can use random names, only offsets have to be the same. offsets aren't copyrightable
    Except random names on a data structure has been proven on court not to remove a copyright claim. The structure is self being the same with evidence you looked at the copyright work to get it can have you screwed.

    Originally posted by pal666 View Post
    would be a problem if someone would need to replicate documentation.
    Yes it is a problem. Its not only when you need to replicate the documentation. Its how did you get your information to reimplement the file system when the information about the implementation be it code or documentation is the same license that is copyleft.

    Originally posted by pal666 View Post
    zero relevance. you can learn format without ever looking at kernel tree: just ask someone else to explain it to you. btw, you are trying to argue that linux kernel is infected by windows license because it implements many windows filesystems. why can't you see how ridiculous it is?
    https://github.com/maharmstone/quibb...END_CMD_SUBVOL

    Why does quibble have LInux kernel ioctl values. Please note they are only in that header file and not used anywhere else in the code base. So we have hand in cookie jar evidence here unfortunately. When something exists in your code base that you don't use that you should not need is direct proof you have been looking at documentation or code you should not have been to write the code. Copyright laws meaning of derived work comes into play here you do not need to have in fact copied 1 word to have copyright infrignment in a derived work claim that was proven in many courts before GPL existed.

    What quibble has done in it header would have been legal of API/ABI was and is declared not copyright-able as it was copying information from a API/ABI. Now if Oracle wins really not so any more.

    The license on that documentation on windows file systems that are in Linux don't have a copyleft license covering the documentation so this problem does not exist. Sorry you arguement is a smoke screen.

    grub2 implementation of btrfs only has what it needs and would have a good chance of arguing fair usage in all cases of copyright claim. Yes the arguement that someone else explained it to the developer could be argued as a possibility in the grub2 case but not in the quibble bootloader case.

    Developers need to be way more careful with these things.

    My recommendation for quibble at this point would be drop their btrfs implementation they have and copy the grub2 one that is gplv3 license as well that is closer to correct or change to gplv2 license while you can as a project gets more developers it can become impossible to change license to correct these problems.

    Leave a comment:


  • profoundWHALE
    replied
    Originally posted by pal666 View Post
    btrfs is aimed to be a replacement for ext4(which is already gpl), there's no zfs on linux. btrfs has more features than zfs, because it was designed later
    Oh please, we both know that btrfs is trying to be like ZFS, but GPL and built with Linux in mind. That's why we see btrfs picking up all these features that ZFS has.

    Originally posted by pal666 View Post
    that't what i'm asking. why spend time on inferior zfs which has bugtracker full of bugs?
    Wait a minute, did you just claim that ZFS is inferior to btrfs. ZFS is a brilliant piece of engineering and btrfs is a mess that is barely holding together.

    Want some evidence? FreeBSD is still alive and kicking mostly due to ZFS. Why? Corporations like their data to not get gobbled up randomly.

    How about companies that specifically offer ZFS with Linux? If btrfs was half as good as you think it is then Canonical would have already made it the default filesystem.A yet ext4 is still the default.

    ​​​​​​​Btrfs isn't good enough to replace ext4 due to inconsistent performance, AND it is too unstable to be used as a replacement for ZFS.

    Originally posted by boxie View Post

    so, I am a longtime user of BTRFS - I even had a bunch of disks in btrfs raid5 (which is rather risky i know)! I have not lost any data and I love the flexibility it gives me for my data storage.
    I'm happy it works well for you. If you happen to get caught in one of it's data destroying bugs like it did to my RAID10 storage then don't say that no one ever warned you.

    Leave a comment:


  • pal666
    replied
    Originally posted by oiaohm View Post
    if Oracle wins API
    here is your problem. filesystem isn't an api. it's data structure. data structures aren't copyrightable. in api you have to use same names, otherwise client code wouldn't compile. in data structure you can use random names, only offsets have to be the same. offsets aren't copyrightable
    Originally posted by oiaohm View Post
    Also the problem is btrfs format documentation is GPL.
    would be a problem if someone would need to replicate documentation.
    Originally posted by oiaohm View Post
    All the documentation on the data format of btrfs is in the Linux kernel source tree.
    zero relevance. you can learn format without ever looking at kernel tree: just ask someone else to explain it to you. btw, you are trying to argue that linux kernel is infected by windows license because it implements many windows filesystems. why can't you see how ridiculous it is?
    Last edited by pal666; 02-14-2020, 04:53 PM.

    Leave a comment:


  • oiaohm
    replied
    Originally posted by pal666 View Post
    well, number of files has zero relevance. what has relevance is "did they use gpl code or just read same data format". and it affects both windows and bootloader equally
    That the problem if Oracle wins API the split you just used might not apply. Also the problem is btrfs format documentation is GPL. Did you use the gpl code or the gpl documentation?? This is the change if Oracle wins over Google we are going to have to watch out for.

    Originally posted by pal666 View Post
    linux code reading/writing btrfs is. on-disk filesystem is a data format, you can patent it, but you can't copyright it. just like linux code reading/writing fat can be affected by patents, but not by windows license
    All the documentation on the data format of btrfs is in the Linux kernel source tree.

    https://yro.slashdot.org/story/20/01...-copyrightable

    Basically you are making the same arguments about data format as what was made about API. Oracle winning that API are copyright-able its only a small jump that data formats are also copyright-able.

    Please note the copyright material causing the taint would be using the data format documentation if Oracle wins the API case. Now if you reverse the data format and don't touch the documentation you would be legal.

    Leave a comment:


  • pal666
    replied
    Originally posted by oiaohm View Post
    Not quite. Is it a derivative work is the question. Kernel module loaded by windows could be argued that its not without any question. In a bootloader it really don't come that simple as the result is a single binary file.
    well, number of files has zero relevance. what has relevance is "did they use gpl code or just read same data format". and it affects both windows and bootloader equally
    Originally posted by oiaohm View Post
    Btrfs is Oracle copyright in the Linux kernel and is only GPL v2.0 no latter clause.
    linux code reading/writing btrfs is. on-disk filesystem is a data format, you can patent it, but you can't copyright it. just like linux code reading/writing fat can be affected by patents, but not by windows license

    Leave a comment:


  • oiaohm
    replied
    Originally posted by pal666 View Post
    it doesn't work that way, otherwise it would require windows to be open source too
    Not quite. Is it a derivative work is the question. Kernel module loaded by windows could be argued that its not without any question. In a bootloader it really don't come that simple as the result is a single binary file.

    Originally posted by pal666 View Post
    that's my question. why filesystem should infect bootloader(or operating system)? it only reads same data format, it doesn't embed linux kernel code
    This will depend on how the Google vs Oracle case goes. Btrfs is Oracle copyright in the Linux kernel and is only GPL v2.0 no latter clause.
    This is something that I hate about the new SPDX.
    // SPDX-License-Identifier: GPL-2.0 is GPL 2.0 Only. As you see all all the btrfs stuff include the documentation.


    // SPDX-License-Identifier: GPL-2.0+
    This is the or latter clause you can use with GPL-3.0 and all the different is having the + so simple to typo and not see it. It was proposed at first to have GPL-2.0-only and GPL-2.0 as the or latter as well so some people remember the proposal not the ratified. The first proposed in my mind made more sense.

    Sorry to say this bootloader might be in some legal trouble.

    Question is does using the Oracle copyrighted Btrfs documentation force you to use GPLv2 only license? This is still in the courts being answered with the Google vs Oracle case. If so this is trouble for grub2 as well.


    Leave a comment:


  • pal666
    replied
    Originally posted by jo-erlend View Post
    What? Why would that be?
    that's my question. why filesystem should infect bootloader(or operating system)? it only reads same data format, it doesn't embed linux kernel code

    Leave a comment:


  • jo-erlend
    replied
    Originally posted by pal666 View Post
    it doesn't work that way, otherwise it would require windows to be open source too
    What? Why would that be?

    Leave a comment:

Working...
X